This study hypothesized that teachers would hold natural expectancies--positive or negative--for students' performance from having previously taught students' older siblings in the same class, depending upon older siblings' performance levels. These students would then differ from no-expectancy controls on specific classroom behaviors and achievement variables. Subjects were students in a Grade 7 English class, all of whom had had an older brother or sister who had been a seventh-grade student in the same building within three previous years. On the basis of grade point averages of older siblings' seventh-grade performance, high-, low-, and medium-expectancy level groups were established. Data was collected through classroom observation using the Brophy-Good Dyadic Interaction System, through language arts achievement testing (Minnesota Achievement Examination), and on teacher-assigned grades. Interpretation of the data indicated high-expectancy experimentals received progressively more, and low-expectancy experimentals much less, teacher praise following correct answers throughout the school year than did their corresponding controls. Achievement testing data showed a considerably lower mean for low-expectancy experimentals than for their corresponding controls, supporting the notion that teachers' negative expectancies increase the probability of students' underachievement. (Author/MJB)